The Research Of Congestion Control Strategies In The Internet

Congestion in the Internet has grow more serious as many new applications have appeared in the Internet these years, such as real time voice and image, which have increased the burden of the Internet load. In order to solve these problems, increasing the capability of the Internet is not enough. There should be sensitive and effective measure to control the congestion. In network communication, congestion will lead to degradation of QoS, so finding an effective way to solve the congestion problem has much significance to improve the performance of the network.Recent research has indicated that end to end flow control of TCP can't satisfy the increasing demand of business in the network, so the network itself should take part in the congestion avoidance and control. The best way is to let IP layer take part in the distribution of resources in the network, that is to say, congestion control mechanism is introduced into the router to supervise and prevent the congestion. As a widely used strategy in congestion control, active queue management (AQM) can control the length of the queue in a certain range effectively to guarantee a preferable throughput and QoS. Random Early Detection (RED) is one of the most effective algorithms of AQM. Its performance is a hotspot in the current research of congestion control.In this paper, the principles of congestion control mechanisms are introduced firstly, and the congestion control in the IP layer is discussed in detail. RED algorithm in the AQM is the focus of this paper. As RED just updated the average queue length when packets arrives, thus the computation shows averaging of packet arrival rate rather than the current buffer occupancy. The dropping of packets will continue for a long time even after the queue has become empty. This will leads to a rapid degradation of the link utilization. So an improved RED algorithm is proposed which combines EWMA with instantaneous queue size for dropping decisions and which also incorporates the adjustments to the EWMA when q < min~th over a sustained period. Simulation studies show that the improved RED algorithm gives a better loss rate and link utilization compared to the original RED algorithms. It is suggested that more studies be done to explore the optimum values of u and ε for different topologies and they should be tied to other RED parameters.